Melanoma Skin Cancer Detection Using Image Processing

Abstract

Among the three basic types of skin cancer, viz, Basal Cell Carcinoma (BCC), Squamous Cell Carcinoma (SCC) and Melanoma, Melanoma is the most dangerous in which survival rate is very low. The proposed skin cancer detection technology is broadly divided into four basic cmponents, viz., image preprocessing which includes hair removal, de-noise, sharpening, resize of the given skin image, segmentation which is used for segmenting out the region of interest from the given image. Here we have used k-means segmentation. The classification algorithm which are going to be used here are Support Vector Machine (SVM).

Introduction

Among three types of skin cancer, viz., Squamous Cell Carcinoma (SCC), Melanoma and Basal Cell Carcinoma (BCC), Melanoma is most dangerous in which survival rate is very low. Early detection of Melanoma can potentially improve survival rate of victim. In USA, in every hour one person dies in melanoma. From a study, it is estimated that around 87,110 new cases of melanoma will be diagnosed in 2018. Among them, 9,730 will die because of melanoma. Melanoma consists of only 1% of all skin cancer cases but the majority of skin cancer death. The vast majority of melanomas are caused by the sun. From a survey done by a UK University, it is found that 86% of melanomas are exposed by ultraviolet (UV) radiation. On average, people’s risk for melanoma doubles if he or she has had more than five sunburns. If a person use SPF 15 or higher SPF sunscreen regularly it can reduce the risk of melanoma by 50% and squamous cell carcinoma by 40%[1].

Save your time!
We can take care of your essay

• Proper editing and formatting
• Free revision, title page, and bibliography
• Flexible prices and money-back guarantee

Place an order

Earlier Work

This is the scenario for which many projects have been tried and developed. Although not same but many related work have been done by many researchers. Some of papers have been referred and explored here. A detailed analysis of the existing systems is done. This study helped in identifying the benefits and also the drawbacks of existing systems.

[1] Enakshi Jana , Dr. Ravi Subban, S.Sarawathi, “Research on skin cancer cell detection using image processing”, 2017. In this paper, an extensive literature survey of current technology is made for skin cancer detection. Of all the methods used for skin cancer detection, SVM and Adaboost produces the best results. [2] Shivangi Jain, Vandana Jagtap, Nitin Pise, “Computer aided melanoma skin cancer detection using Image processing”, 2015. In this paper it is concluded that the proposed system can be effectively used by patients and physicians to diagnose the skin cancer more accurately. This tool is more useful for the rural areas where the experts in the medical field may not be available. [3] Vijayalkshmi M.M, “Melanoma skin cancer detection using Image processing and machine learning”, June 2019. The aim of this project is to determine the accurate prediction of skin cancer and also to classify the skin cancer as malignant or non-malignant melanoma. [4] Sanjana M, Dr. V. Hanuman Kumar, “Skin cancer detection using Machine learning algorithm”, Dec 2018. This paper focuses on determining the stage of the skin cancer, based on various feature such as the area of the spread, diameter, color of the lesion, etc. The analysis can be made with the help of machine learning algorithm.[5] Jianpeng Qi, Yanwei Yu*, Lihong Wang,and Jinglei Liu School of Computer and Control Engineering, Yantai University, Yantai, Shandong 264005, China, “K*-Means: An Effective and Efficient K-means Clustering Algorithm”, in 2016 IEEE International Conferences on Big Data and Cloud Computing (BDCloud), Social Computing and Networking (SocialCom), Sustainable Computing and Communications (SustainCom). [6]Ginu George, 2 Rinoy Mathew Oommen, 3 Shani Shelly, 4 Stephie Sara Philipose, 5 Ann Mary Varghese 1,2,3 UG Scholar, Department of Computer Science and Engineering, “A Survey on Various Median Filtering Techniques For Removal of Impulse Noise From Digital Image”, IEEE Conference on Emerging Devices and Smart Systems (ICEDSS 2018) 2-3 March 2018.

The system has two parts, training and testing. Both parts undergo following steps.

• Step 1] Our first step involves the collection of Images.For this purpose we have collected 5026 Dermoscopy Images taken from the dermoscope.
• Step 2]In the second step we are doing the image preprocessing. The process involved in the preprocessing are like hair removal, noise removal, enhancement, colour conversion. In hair removal algorithm we have used opening and closing algorithm for noise removal we have used median filter.For enhancement purpose we have used histogram. For training our model we needed all the images of the same dimension, so for that purpose we resised the images.
• Step 3] Third step is segmentation using k-mean clustering, , k-means heavily depends on the position of initial centers, and the chosen starting centers randomly may lead to poor quality of clustering. In our paper we have optimized k-means clustering method along with three optimization principles named k∗-means.

Segmentation is followed by feature extraction. No machine learning algorithm can work without predefined features set. The type of features can be broadly divided into following categories.

Shape Features - Asymmetry, Compactness, Ulnar Variance, Diameters. 2.Texture Features - GLCM (Gray-Level CoOccurrence Matrix), Coarseness. 3. Color Features - Variance, Entropy, Skewness. 4. Contrast - Measure of the local variations and texture of shadow depth. 5. Homogeneity - Measure of closeness of the distribution of elements. The feature of nucleus is extracted using WAVELET TRANSFORM and GLCM.

• Step 4] In training part features of pure cancer cell is stored in data base.

In the testing part, the cell which needs to be tested is taken as input.

• Step 5] Finally SVM classifier with the help of data in the data base is used for classification, where decision is done whether the cell is cancerous or not.

CONCLUSION

The proposed system of skin cancer detection can be implemented using support vector machine to classify easily whether image is cancerous or non-cancerous. The system will determine the stage of the skin cancer, based on various features such as the area of the spread, diameter, color of the lesion, etc. The analysis can be made with the help of the machine learning algorithm, in which we train the system based on the history of the images stored in the database, and the test image comes in the category of the melanoma or not, if it does, then to determine its stage. A comparison can be made with the existing systems, machine learning reduces the computational time. Hence, the treatment can begin faster.

REFERENCES

1. Enakshi Jana , Dr. Ravi Subban, S.Sarawathi, “Research on skin cancer cell detection using image processing”, 2017.
2. Shivangi Jain, Vandana Jagtap, Nitin Pise, “Computer aided melanoma skin cancer detection using Image processing”, 2015.
3. Vijayalkshmi M.M, “Melanoma skin cancer detection using Image processing and machine learning”, June 2019.
4. Sanjana M, Dr. V. Hanuman Kumar, “Skin cancer detection using Machine learning algorithm”, Dec 2018.
5. Jianpeng Qi, Yanwei Yu*, Lihong Wang,and Jinglei Liu School of Computer and Control Engineering, Yantai University, Yantai, Shandong 264005, China, “K*-Means: An Effective and Efficient K-means Clustering Algorithm”, in 2016 IEEE International Conferences on Big Data and Cloud Computing (BDCloud), Social Computing and Networking (SocialCom), Sustainable Computing and Communications (SustainCom).
6. Ginu George, 2 Rinoy Mathew Oommen, 3 Shani Shelly, 4 Stephie Sara Philipose, 5 Ann Mary Varghese 1,2,3 UG Scholar, Department of Computer Science and Engineering, “A Survey on Various Median Filtering Techniques For Removal of Impulse Noise From Digital Image”, IEEE Conference on Emerging Devices and Smart Systems (ICEDSS 2018) 2-3 March 2018.
7. https://medium.com/machine-learning-101/chapter-2-svm-support-vector-machine theory-f0812effc72
8. https://towardsdatascience.com/https-medium-com-pupalerushikesh-svm-f4b42800e989
9. Pratik Dubal, Sankirtan Bhat, Chaitanya Joglekar, Dr. Sonal Patil, “Skin cancer detection and classification”, 2017.
10. Mohd Afizi Mohd Shukran, Nor Suraya Mariam Ahmad, Farahana Rahmat, “Melanoma cancer diagnosis device using Image processing techniques”, Feb 2019.
11. Nay Chi lynn, Zin Mar Kyu, “Segmentation and classification of skin cancer melanoma from skin lesion images”, 2017.